Scholarlink

<p>Identifying influential scholars within academic ecosystems is a central challenge in socionetwork analysis and the design of intelligent knowledge platforms. Traditional bibliometric indices often overlook the structural dynamics of collaboration networks, leading to biased or incomplete assessments of scholarly influence. This study introduces a socionetwork-based evaluation framework that analyzes 22 classical and advanced centrality measures within large-scale co-authorship networks. Using a hybrid sampling strategy that integrates K-means clustering, silhouette scoring, and tournament selection, we address class imbalance in scholar recognition tasks. Results show that Load, PageRank, Betweenness, and Degree centrality measures most effectively capture influence patterns relevant to academic ranking. Additionally, inter-centrality correlation analysis reveals distinct strategic dimensions of scholarly recognition, offering insights for developing transparent, structure-aware ranking engines in academic information systems. The findings contribute to network-based evaluation methodologies applicable to broader socio-technical and organizational contexts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Socionetwork-Based Evaluation of Academic Systems: Integrating Centrality Metrics into Empirical Ranking Frameworks

  • 期刊论文

  • 发表日期：

    2026年2月14日

查看全文
  • Amer Zaheer,
  • Ghulam Mustafa,
  • Muhammad Tanvir Afzal

摘要

Identifying influential scholars within academic ecosystems is a central challenge in socionetwork analysis and the design of intelligent knowledge platforms. Traditional bibliometric indices often overlook the structural dynamics of collaboration networks, leading to biased or incomplete assessments of scholarly influence. This study introduces a socionetwork-based evaluation framework that analyzes 22 classical and advanced centrality measures within large-scale co-authorship networks. Using a hybrid sampling strategy that integrates K-means clustering, silhouette scoring, and tournament selection, we address class imbalance in scholar recognition tasks. Results show that Load, PageRank, Betweenness, and Degree centrality measures most effectively capture influence patterns relevant to academic ranking. Additionally, inter-centrality correlation analysis reveals distinct strategic dimensions of scholarly recognition, offering insights for developing transparent, structure-aware ranking engines in academic information systems. The findings contribute to network-based evaluation methodologies applicable to broader socio-technical and organizational contexts.